<ul><li>Discovering materials with desirable properties in an efficient way remains a significant problem in materials science.</li><li>LLM-Fusion is a novel multimodal fusion model that integrates diverse representations for accurate material property prediction.</li><li>The model leverages large language models (LLMs) to combine different sources of information, such as SMILES, SELFIES, text descriptions, and molecular fingerprints.</li><li>LLM-Fusion outperforms traditional methods in terms of accuracy and shows promising results on multiple prediction tasks.</li></ul>

LLM-Fusion: A Novel Multimodal Fusion Model for Accelerated Material Discovery

Discover more